Despite the recent advances in Large Language Models, benchmarks for evaluating legal writing remain scarce due to the inherent complexity of assessing open-ended responses in this domain. One of the key challenges in evaluating language models on domain-specific tasks is finding test datasets that are public, frequently updated, and contain comprehensive evaluation guidelines. The Brazilian Bar Examination meets these requirements. We introduce oab-bench, a benchmark comprising 105 questions across seven areas of law from recent editions of the exam. The benchmark includes comprehensive evaluation guidelines and reference materials used by human examiners to ensure consistent grading. We evaluate the performance of four LLMs on oab-bench, finding that Claude-3.5 Sonnet achieves the best results with an average score of 7.93 out of 10, passing all 21 exams. We also investigated whether LLMs can serve as reliable automated judges for evaluating legal writing. Our experiments show that frontier models like OpenAI’s o1 achieve a strong correlation with human scores when evaluating approved exams, suggesting their potential as reliable automated evaluators despite the inherently subjective nature of legal writing assessment. The source code1 and the benchmark2 – containing questions, evaluation guidelines, model-generated responses, and their respective automated evaluations – are publicly available.

Leandro De Souza, Thales Almeida, Roberto Lotufo, Rodrigo Frassetto Nogueira. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.

In recent years, Augmentative and Alternative Communication (AAC) systems have grown significantly in Brazil, particularly for individuals with cognitive disorders who rely on high-tech AAC tools. Artificial Intelligence (AI) has significantly improved high-tech AAC systems by enhancing accessibility, increasing output generation speed, and improving AAC interfaces' customization and adaptability. This study investigates the use of Large Language Models (LLMs) to generate synthetic text data to augment a corpus for AAC in Brazilian Portuguese. A three-step method was used to augment an initial corpus of 667 AAC-like sentences produced by specialists to a corpus of 13k sentences, comprising sentence collection, corpus augmentation using GPT-3 in a few-shot setting, and corpus cleaning. The quality and reliability of the generated corpus were assessed through a coverage analysis, comparing the content of the generated sentences with the original human-composed sentences. The results provide insights into the methods' strengths and limitations and inform future efforts to improve the generation of synthetic text data for the AAC domain in Brazilian Portuguese.

Due to the high volume of cases in transit in the Brazilian judicial system, it is of great interest to automate processes that help legal professionals. With the recent advances in text generation methods, there is potential to apply these techniques in tasks that require automation. In this article, we address the problem of automating the writing of keyphrases, a sequence of key terms present in documents used in courts throughout Brazil. For this, we are proposing the use of a text-to-text framework based on generative Transformers. We evaluated several generative Transformer models (such as PTT5, mT5, OPT, and BLOOM) for the proposal, and a comparison of their performance is presented. The best-performing model, PTT5, was adopted as the text generator since it achieved a BLEU score of 37.54% in the test set, outperforming other evaluated methods by up to 24.6%. Finally, to assess the influence of keyphrases and the quality of the generated ones, we performed several experiments based on a real case of legal information retrieval. By using traditional information retrieval methods, such as TF-IDF and BM25, in combination with the original, generated keyphrases, or both, we observed gains statistically significant (p-value${<}$0.05) in all experiments.

This tutorial will provide an overview of recent advances on neuro-symbolic approaches for information retrieval. A decade ago, knowledge graphs and semantic annotations technology led to active research on how to best leverage symbolic knowledge. At the same time, neural methods have demonstrated to be versatile and highly effective.

Recent work has shown that incorporating explanations into the output generated by large language models (LLMs) can significantly enhance performance on a broad spectrum of reasoning tasks. Our study extends these findings by demonstrating the benefits of explanations for neural rankers. By utilizing LLMs such as GPT-3.5 to enrich retrieval datasets with explanations, we trained a sequence-to-sequence ranking model, dubbed ExaRanker, to generate relevance labels and explanations for query-document pairs. The ExaRanker model, finetuned on a limited number of examples and synthetic explanations, exhibits performance comparable to models finetuned on three times more examples, but without explanations. Moreover, incorporating explanations imposes no additional computational overhead into the reranking step and allows for on-demand explanation generation. The codebase and datasets used in this study will be available at https://github.com/unicamp-dl/ExaRanker

Developing a universal model that can efficiently and effectively respond to a wide range of information access requests-from retrieval to recommendation to question answering---has been a long-lasting goal in the information retrieval community. This paper argues that the flexibility, efficiency, and effectiveness brought by the recent development in dense retrieval and approximate nearest neighbor search have smoothed the path towards achieving this goal. We develop a generic and extensible dense retrieval framework, called framework, that can handle a wide range of (personalized) information access requests, such as keyword search, query by example, and complementary item recommendation. Our proposed approach extends the capabilities of dense retrieval models for ad-hoc retrieval tasks by incorporating user-specific preferences through the development of a personalized attentive network. This allows for a more tailored and accurate personalized information access experience. Our experiments on real-world e-commerce data suggest the feasibility of developing universal information access models by demonstrating significant improvements even compared to competitive baselines specifically developed for each of these individual information access tasks. This work opens up a number of fundamental research directions for future exploration.

The zero-shot cross-lingual ability of models pretrained on multilingual and even monolingual corpora has spurred many hypotheses to explain this intriguing empirical result. However, due to the costs of pretraining, most research uses public models whose pretraining methodology, such as the choice of tokenization, corpus size, and computational budget, might differ drastically. When researchers pretrain their own models, they often do so under a constrained budget, and the resulting models might underperform significantly compared to SOTA models. These experimental differences led to various inconsistent conclusions about the nature of the cross-lingual ability of these models. To help further research on the topic, we released 10 monolingual byte-level models rigorously pretrained under the same configuration with a large compute budget (equivalent to 420 days on a V100) and corpora that are 4 times larger than the original BERT’s. Because they are tokenizer-free, the problem of unseen token embeddings is eliminated, thus allowing researchers to try a wider range of cross-lingual experiments in languages with different scripts. Additionally, we release two models pretrained on non-natural language texts that can be used in sanity-check experiments. Experiments on QA and NLI tasks show that our monolingual models achieve competitive performance to the multilingual one, and hence can be served to strengthen our understanding of cross-lingual transferability in language models.

The Information Retrieval (IR) community has recently witnessed a revolution due to large pretrained transformer models. Another key ingredient for this revolution was the MS MARCO dataset, whose scale and diversity has enabled zero-shot transfer learning to various tasks. However, not all IR tasks and domains can benefit from one single dataset equally. Extensive research in various NLP tasks has shown that using domain-specific training data, as opposed to a general-purpose one, improves the performance of neural models. In this work, we harness the few-shot capabilities of large pretrained language models as synthetic data generators for IR tasks. We show that models finetuned solely on our synthetic datasets outperform strong baselines such as BM25 as well as recently proposed self-supervised dense retrieval methods. Code, models, and data are available at https://github.com/zetaalphavector/inpars.

With doc2query, we train a neural sequence-to-sequence model that, given an input span of text, predicts a natural language query that the text might answer. These predictions can be viewed as document expansions that feed standard bag-of-words term weighting models such as BM25 or neural retrieval models based on learned sparse lexical representations such as uniCOIL. Previous experiments on the MS MARCO datasets have demonstrated the effectiveness of these methods, and they serve as baselines that are widely used by the community today. Following the recent release of the MS MARCO V2 passage and document ranking test collections, we have refreshed our doc2query and uniCOIL models. This work describes a number of resources that support competitive, reproducible baselines for both the MS MARCO V1 and V2 test collections using our Anserini and Pyserini IR toolkits. Together, they provide a solid foundation for future research on neural retrieval models using the MS MARCO datasets and beyond.

There has been mounting evidence that pretrained language models fine-tuned on large and diverse supervised datasets can transfer well to a variety of out-of-domain tasks. In this work, we investigate this transfer ability to the legal domain. For that, we participated in the legal case entailment task of COLIEE 2021, in which we use such models with no adaptations to the target domain. Our submissions achieved the highest scores, surpassing the second-best submission by more than six percentage points. Our experiments confirm a counter-intuitive result in the new paradigm of pretrained language models: given limited labeled data, models with little or no adaption to the target task can be more robust to changes in the data distribution than models fine-tuned on it. Code is available at https://github.com/neuralmind-ai/coliee.

Pyserini is a Python toolkit for reproducible information retrieval research with sparse and dense representations. It aims to provide effective, reproducible, and easy-to-use first-stage retrieval in a multi-stage ranking architecture. Our toolkit is self-contained as a standard Python package and comes with queries, relevance judgments, pre-built indexes, and evaluation scripts for many commonly used IR test collections. We aim to support, out of the box, the entire research lifecycle of efforts aimed at improving ranking with modern neural approaches. In particular, Pyserini supports sparse retrieval (e.g., BM25 scoring using bag-of-words representations), dense retrieval (e.g., nearest-neighbor search on transformer-encoded representations), as well as hybrid retrieval that integrates both approaches. This paper provides an overview of toolkit features and presents empirical results that illustrate its effectiveness on two popular ranking tasks. Around this toolkit, our group has built a culture of reproducibility through shared norms and tools that enable rigorous automated testing.

The COVID-19 pandemic has brought about a proliferation of harmful news articles online, with sources lacking credibility and misrepresenting scientific facts. Misinformation has real consequences for consumer health search, i.e., users searching for health information. In the context of multi-stage ranking architectures, there has been little work exploring whether they prioritize correct and credible information over misinformation. We find that, indeed, training models on standard relevance ranking datasets like MS MARCO passage---which have been curated to contain mostly credible information---yields models that might also promote harmful misinformation. To rectify this, we propose a label prediction technique that can separate helpful from harmful content. Our design leverages pretrained sequence-to-sequence transformer models for both relevance ranking and label prediction. Evaluated at the TREC 2020 Health Misinformation Track, our techniques represent the top-ranked system: Our best submitted run was 19.2 points higher than the second-best run based on the primary metric, a 68% relative improvement. Additional post-hoc experiments show that we can boost effectiveness by another 3.5 points.

The goal of text ranking is to generate an ordered list of texts retrieved from a corpus in response to a query. Although the most common formulation of text ranking is search, instances of the task can also be found in many natural language processing applications. This tutorial, based on a forthcoming book, provides an overview of text ranking with neural network architectures known as transformers, of which BERT is the best-known example. The combination of transformers and self-supervised pretraining has, without exaggeration, revolutionized the fields of natural language processing (NLP), information retrieval (IR), and beyond. We provide a synthesis of existing work as a single point of entry for both researchers and practitioners. Our coverage is grouped into two categories: transformer models that perform reranking in multi-stage ranking architectures and learned dense representations that perform ranking directly. Two themes pervade our treatment: techniques for handling long documents and techniques for addressing the tradeoff between effectiveness (result quality) and efficiency (query latency). Although transformer architectures and pretraining techniques are recent innovations, many aspects of their application are well understood. Nevertheless, there remain many open research questions, and thus in addition to laying out the foundations of pretrained transformers for text ranking, we also attempt to prognosticate the future.

Chatty Goose is an open-source Python conversational search framework that provides strong, reproducible reranking pipelines built on recent advances in neural models. The framework comprises extensible modular components that integrate with popular libraries such as Transformers by HuggingFace and ParlAI by Facebook. Our aim is to lower the barrier of entry for research in conversational search by providing reproducible baselines that researchers can build on top of. We provide an overview of the framework and demonstrate how to instantiate a new system from scratch. Chatty Goose incorporates improvements to components that we introduced in the TREC 2019 Conversational Assistance Track (CAsT), where our submission represented the top-performing system. Using our framework, a comparable run can be reproduced with just a few lines of code.

The goal of text ranking is to generate an ordered list of texts retrieved from a corpus in response to a query. Although the most common formulation of text ranking is search, instances of the task can also be found in many natural language processing applications. This tutorial, based on a forthcoming book, provides an overview of text ranking with neural network architectures known as transformers, of which BERT is the best-known example. The combination of transformers and self-supervised pretraining has, without exaggeration, revolutionized the fields of natural language processing (NLP), information retrieval (IR), and beyond. We provide a synthesis of existing work as a single point of entry for both researchers and practitioners. Our coverage is grouped into two categories: transformer models that perform reranking in multi-stage ranking architectures and learned dense representations that perform ranking directly. Two themes pervade our treatment: techniques for handling long documents and techniques for addressing the tradeoff between effectiveness (result quality) and efficiency (query latency). Although transformer architectures and pretraining techniques are recent innovations, many aspects of their application are well understood. Nevertheless, there remain many open research questions, and thus in addition to laying out the foundations of pretrained transformers for text ranking, we also attempt to prognosticate the future.

Conversational search plays a vital role in conversational information seeking. As queries in information seeking dialogues are ambiguous for traditional ad hoc information retrieval (IR) systems due to the coreference and omission resolution problems inherent in natural language dialogue, resolving these ambiguities is crucial. In this article, we tackle conversational passage retrieval, an important component of conversational search, by addressing query ambiguities with query reformulation integrated into a multi-stage ad hoc IR system. Specifically, we propose two conversational query reformulation (CQR) methods: (1) term importance estimation and (2) neural query rewriting. For the former, we expand conversational queries using important terms extracted from the conversational context with frequency-based signals. For the latter, we reformulate conversational queries into natural, stand-alone, human-understandable queries with a pretrained sequence-to-sequence model. Detailed analyses of the two CQR methods are provided quantitatively and qualitatively, explaining their advantages, disadvantages, and distinct behaviors. Moreover, to leverage the strengths of both CQR methods, we propose combining their output with reciprocal rank fusion, yielding state-of-the-art retrieval effectiveness, 30% improvement in terms of NDCG@3 compared to the best submission of Text REtrieval Conference (TREC) Conversational Assistant Track (CAsT) 2019.

While internalized "implicit knowledge" in pretrained transformers has led to fruitful progress in many natural language understanding tasks, how to most effectively elicit such knowledge remains an open question.Based on the text-to-text transfer transformer (T5) model, this work explores a template-based approach to extract implicit knowledge for commonsense reasoning on multiple-choice (MC) question answering tasks.Experiments on three representative MC datasets show the surprisingly good performance of our simple template, coupled with a logit normalization technique for disambiguation.Furthermore, we verify that our proposed template can be easily extended to other MC tasks with contexts such as supporting facts in open-book question answering settings.Starting from the MC task, this work initiates further research to find generic natural language templates that can effectively leverage stored knowledge in pretrained models.

Search engines play an important role in our everyday lives by assisting us in finding the information we need.When we input a complex query, however, results are often far from satisfactory.In this work, we introduce a query reformulation system based on a neural network that rewrites a query to maximize the number of relevant documents returned.We train this neural network with reinforcement learning.The actions correspond to selecting terms to build a reformulated query, and the reward is the document recall.We evaluate our approach on three datasets against strong baselines and show a relative improvement of 5-20% in terms of recall.Furthermore, we present a simple method to estimate a conservative upperbound performance of a model in a particular environment and verify that there is still large room for improvements.

We propose a goal-driven web navigation as a benchmark task for evaluating an agent with abilities to understand natural language and plan on partially observed environments. In this challenging task, an agent navigates through a website, which is represented as a graph consisting of web pages as nodes and hyperlinks as directed edges, to find a web page in which a query appears. The agent is required to have sophisticated high-level reasoning based on natural languages and efficient sequential decision-making capability to succeed. We release a software tool, called WebNav, that automatically transforms a website into this goal-driven web navigation task, and as an example, we make WikiNav, a dataset constructed from the English Wikipedia. We extensively evaluate different variants of neural net based artificial agents on WikiNav and observe that the proposed goal-driven web navigation well reflects the advances in models, making it a suitable benchmark for evaluating future progress. Furthermore, we extend the WikiNav with question-answer pairs from Jeopardy! and test the proposed agent based on recurrent neural networks against strong inverted index based search engines. The artificial agents trained on WikiNav outperforms the engined based approaches, demonstrating the capability of the proposed goal-driven navigation as a good proxy for measuring the progress in real-world tasks such as focused crawling and question-answering.

With the growing use of biometric authentication systems in the recent years, spoof fingerprint detection has become increasingly important. In this paper, we use convolutional neural networks (CNNs) for fingerprint liveness detection. Our system is evaluated on the data sets used in the liveness detection competition of the years 2009, 2011, and 2013, which comprises almost 50 000 real and fake fingerprints images. We compare four different models: two CNNs pretrained on natural images and fine-tuned with the fingerprint images, CNN with random weights, and a classical local binary pattern approach. We show that pretrained CNNs can yield the state-of-the-art results with no need for architecture or hyperparameter selection. Data set augmentation is used to increase the classifiers performance, not only for deep architectures but also for shallow ones. We also report good accuracy on very small training sets (400 samples) using these large pretrained networks. Our best model achieves an overall rate of 97.1% of correctly classified samples-a relative improvement of 16% in test error when compared with the best previously published results. This model won the first prize in the fingerprint liveness detection competition 2015 with an overall accuracy of 95.5%.